Variable mechanical damping in turbogenerator torsional oscillations
The prediction of turbogenerator behaviour following fault or disturbance conditions has attracted much interest over the past decade, due to the large amplitude torsional oscillations which can be induced. Many simulation studies have been conducted, but the majority have used d-q models to simulate the electrical machines, and have modelled the turbine shaft damping as constant viscous damping. The d-q machine representation is adequate for balanced conditions, whereas the phase coordinate electrical machine model may be used to simulate unbalanced operation. The latter is adopted for the present work. Constant viscous shaft damping is known to be an over simplification which does not truly reflect the stress level dependent, frequency independent nature of shaft material damping. Although small, the mechanical damping can have a decisive influence over the rate of decay of torsional oscillations, and hence on shaft fatigue life expenditure. This project seeks principally to enhance the shaft torsional model formulation by developing methods of simulating variable shaft damping. Two methods are proposed; one in which a stress-strain hysteresis loop is simulated for each shaft section leading to a variable shaft stiffness, and a second simpler method, in which the viscous damping representation is retained, but with variable viscous damping coefficients. It is shown that the effect of incorporating variable damping is to increase the rate of decay of the initial post-fault oscillations, but then to prolong the subsequent oscillations. This is the effect generally observed in practice on full size turbogenerators. Experimental results from a small single shaft rig also show this effect and agree closely with the predictions of the theory. A brief study of the fatigue life expenditure in a turbogenerator shaft during post-fault oscillations is also presented. It is concluded that both the proposed methods of simulating variable mechanical damping have feasible practical applications, and that the use of variable damping in simulation models can significantly improve the prediction of rate of decay of turbogenerator post-fault torsional oscillations.